2.01-1 Deconstructing the New Zealand Transport Agency Economic Evaluation Manual
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Deconstructing the New Zealand
Transport Agency Economic
Evaluation Manual:
Bringing the best research on transportation and its economic benefit into the
applied New Zealand context.
2.01—1
MAPP575 – Dissertation
Final Draft
Due December 2013
Handed in September 2013
Thomas S. Pettit
Master of Public Policy Candidate
Student ID – 300274621
Supervisor – Toby Daglish
2.01—2
Abstract
This paper explores the New Zealand Transport Agency Economic Evaluation
Manual from the perspective of best practices in international literature. Drawing upon
research from the international community and policy-focused bodies like the OECD, the
paper seeks to improve the NZTA EEM’s quality by employing hedonic and revealed-
preference methods to create a more accurate tool to derive the value of certain transport
investments. The paper finds that the value of time in New Zealand is far too low, the
discount regime improperly reflects the nature of such investments, cycling benefits are
undervalued, and property values are not accounted for as well as they could be. The
paper then applies these findings to the recently-completed Public Transport Spine Study
in Wellington, New Zealand to illustrate the importance of accurate economic evaluation
of transport investments.
Acknowledgements:
My deepest thanks to Toby and Claudia for their advice, and Nadine for her support.
2.01—3
Table Of Contents
Table Of Contents....................................................................... 2.01—4
Index of Tables, Equations and Figures ......................................... 2.01—5
Chapter I. Introduction ................................................................. I—6
Chapter II. Existing New Zealand Context, Recent Research, and Optimal
CBA Methods ................................................................................ II—7
Section 2.01 The Discount Rate ........................................................................................2.01—7
Section 2.02 The Value of Time ......................................................................................2.02—12
Section 2.03 Treatment of Cycling Benefits.....................................................................2.03—16
Section 2.04 Patterns of Development and Transport-Inspired Value............................2.04—18
Section 2.05 Automobile Ownership Costs......................................................................2.05—20
Section 2.06 The Community Severance (Barrier) Effect................................................2.06—21
Section 2.07 Summary .....................................................................................................2.07—23
Chapter III. Spine Study Input Analysis....................................... III—23
Section 3.01 The Travel Times........................................................................................3.01—23
Section 3.02 The Ridership..............................................................................................3.02—26
Section 3.03 Properties Receiving Value Uplift...............................................................3.03—28
Section 3.04 Difference in Car Kilometers Travelled......................................................3.04—29
Chapter IV. Methodology ...........................................................IV—31
Section 4.01 Congestion Times........................................................................................4.01—31
Section 4.02 Travel Time Savings ...................................................................................4.02—33
Section 4.03 Automobile Ownership Costs......................................................................4.03—34
Section 4.04 Community Severance ................................................................................4.04—35
Chapter V. Aggregate Results....................................................... V—36
Section 5.01 Corrected Travel Times and International Costs .......................................5.01—36
Value of Property Uplift (in Millions of Dollars) .........................................................................5.01—36
Section 5.02 Corrected Travel Times and Current NZTA Costs ...................................5.02—37
Value of Property Uplift (in Millions of Dollars) .........................................................................5.02—37
Section 5.03 Spine Study Travel Times and International Costs ....................................5.03—39
Value of Property Uplift (in Millions of Dollars) .........................................................................5.03—39
Section 5.04 Existing Analysis ..........................................................................................5.04—39
Value of Property Uplift (in Millions of Dollars) .........................................................................5.04—40
Chapter VI. Discussion ..............................................................VI—41
Section 6.01 The Discount Rate ......................................................................................6.01—41
Section 6.02 The Value of Time ......................................................................................6.02—42
Section 6.03 Automobile Ownership Costs......................................................................6.03—42
Section 6.04 Community Severance Effect......................................................................6.04—43
Section 6.05 Property Value Uplift ..................................................................................6.05—43
Section 6.06 Ridership and Congestion Times ................................................................6.06—44
Chapter VII. Conclusions ..........................................................VII—44
International Implications............................................................................................................VII—46
Bibliography ..............................................................................VII—48
APPENDIX ................................................................................VII—58
2.01—4
Appendix A – Congestion Times..........................................................................................VII—58 Appendix B – Congestion Times without Flyover & 2nd Tunnel .........................................VII—59 Appendix C – Costs and Benefits with No SH1 Improvements...........................................VII—61 Appendix D – Perturbation Analysis ....................................................................................VII—61 Appendix E – Cost Structure When Discounted..................................................................VII—65 Appendix F – Financial Recovery ........................................................................................VII—65 Appendix G - Realistic Cost Structures................................................................................VII—67 Appendix H - Benefit-Cost Ratios........................................................................................VII—67 Index of Tables, Equations and Figures Table II-1 Existing International & Proposed NZ Discounting Schemes ____________________________ 2.01—10 Table II-2 Proposed Changes to Cycling Benefits ____________________________________________ 2.03—18 Table III-1 Congestion Times Through the Basin Reserve_______________________________________ 3.01—24 Table V-1 Costs and Benefits (in Millions) with Corrected Data and Values _________________________ 5.01—36 Table V-2 Value of Property Uplift - Expected ______________________________________________ 5.01—36 Table V-3 Costs and Benefits (in Millions) with Corrected Data and Current Values ____________________ 5.02—37 Table V-4 Value of Property Uplift - Expected _____________________________________________ 5.02—37 Table V-5 Costs and Benefits (in Millions) with Current Data and Corrected Values____________________ 5.03—39 Table V-6 Value of Property Uplift - Expected ______________________________________________ 5.03—39 Table V-7 Costs and Benefits (in Millions) with Current Data and Values __________________________ 5.04—40 Table V-8 Value of Property Uplift - Possible_______________________________________________ 5.04—40 Table A-0-1 Congestion Times without a Flyover or 2nd Tunnel ___________________________________VII—59 Table A0-2 Costs and Benefits (in Millions) with Corrected Data/Values – No SH1 Improvements___________VII—61 Table A-0-3 Costs and Benefits (in Millions) - Perturbed Value of Time_____________________________VII—61 Table A-0-4 Costs and Benefits (in Millions) - Perturbed Congestion _______________________________VII—62 Table A-0-5 Costs and Benefits (in Millions) - Perturbed Ridership Growth __________________________VII—63 Table A-0-6 10 Year Bond Cost Structure – Interest + Social Discounting Applied ______________________VII—65 Table A-0-7 Financial Recoveries and Costs ________________________________________________VII—66 Equation 1 – Use Ratio _____________________________________________________________ 4.01—32 Equation 2 – Congestion Parameters _____________________________________________________ 4.01—32 Equation 3 – Total Time Savings_______________________________________________________ 4.02—33 Equation 4 – Time Savings Per Rider ____________________________________________________ 4.02—33 Equation 5 – Net Travel Time Savings ___________________________________________________ 4.02—34 Equation 6 - Auto Ownership Savings ___________________________________________________ 4.03—34 Equation 7 – Community Severance Costs(or Benefits) _________________________________________ 4.04—35 Figure 1 – Properties affected by Value Uplift _______________________________________________ 3.03—28 2.01—5
Chapter I. Introduction
In modern political decision making cost-benefit analysis is the most common,
most powerful, and most important tool employed to determine whether or not a public
project should be executed. The cost-benefit methodology currently in use in New
Zealand is in need of a review, particularly in the transport area. The transport methods
stand out as having areas that need better research behind them if the best evidence is to
reach the hands of decision makers. Whether it is fundamental issues that span disciplines
like the discount rate, critical elements of transportation analysis like the value of time, or
unique changes relative to individual transportation modes like cycling and driving, there
are many issues that currently prevent cost benefit analysis in New Zealand from
providing the best possible evidence.
Characteristic of these shortcomings is the Public Transport Spine Study recently
released for Wellington, New Zealand: an ordinary case where a cost-benefit analysis was
produced to analyze an engineering study according to the methods of the New Zealand
Transport Agency Economic Evaluation Manual (NZTA EEM). This case illustrates
what these issues can lead to – undervalued benefits, costs and dissonant values between
modes. Despite this, the study contributes to the study of Wellington’s transport context –
particularly relative to the costs of construction and the engineering challenges faced by
each mode. However, the conclusions of the report have important implications for
Wellington’s public transport infrastructure, and those conclusions are built upon faulty
foundations.
This paper examines 3 aspects of cost-benefit analysis. Firstly, the NZTA EEM is
dissected relative to six challenges this paper perceives: discounting practice, value of time
determination, cycling safety and health benefits, employing land values as a confirming
value of benefits, excluding private automobile ownership costs from initial benefit
calculations, and neglecting to determine the value of the community severance effect.
This paper compares current methods in each of these categories to international best
practice in order to determine what the difference between current practice and best
practice may be. This results in a package of best practice methods that will then be
applied later in the paper.
The second part of this paper explains these issues in the context of the Public
Transport Spine Study. This requires a two-step process. In order to be sure of this
paper’s conclusions, the Spine Study must be scrutinized if it is to serve as the foundation
for conclusions. This is particularly true given that the study evaluates modal choices not
currently extant to study within a New Zealand context, and so conclusions about cost-
benefit inputs must be scrutinized much the same way this paper examines the
methodology used to create cost-benefit outputs. This paper restricts analysis to travel
times anticipated given future projected car growth and ridership projections per mode.
This is because these two factors are by far the most important for determining overall
benefits of a given option; even more important in some ways than the cost benefit
analysis methodology applied to them.
I—6
Finally, the third portion takes the inputs, and then runs through the cost-benefit
analysis practice put forth in each of the relevant areas, in order to provide an analysis of
four separate possible scenarios:
1. The existing benefits and costs determined by the Spine Study
2. The benefits and costs determined using Spine Study travel times and ridership
data given corrected costs.
3. The benefits and costs determined using corrected travel times and ridership data,
as well as corrected costs.
4. The benefits and costs determined using corrected travel times and ridership data,
as well as corrected costs presuming a flyover and 2nd Mount Victoria Tunnel are
not constructed.
Through the analysis of these four scenarios this paper can determine to what degree
existing cost-benefit analysis is a problem and where corrections to that problem may be
sought soonest.
This paper seeks to reduce the degree of error in New Zealand cost-benefit
analysis, particularly for the transport sector. In this vein, the paper seeks better
information for decision makers, better projects getting funded, better options for
ratepayers, and better long-term outcomes for the country.
Chapter II. Existing New Zealand Context, Recent
Research, and Optimal CBA Methods
Transport investment decisions lean heavily on cost-benefit analysis to determine
their viability. In turn, these analyses depend heavily on the parameters employed to
evaluate them. The paper identifies six areas where current New Zealand methodology
does not conform with international literature: discounting; the value of time; cycling
benefit calculation; property uplift; automobile ownership costs; and community
severance. Each one deserves close attention, and improving the methodology used to
calculate these components of cost-benefit analyses would improve the evidence provided
to decisionmakers and ideally improve outcome attainment for transport investments.
This improvement will result in better internalizing externalities and helping to seek a
more balanced, realistic version of cost-benefit analysis. In doing so, these changes strike a
compromise between the myriad needs of decisionmakers while attempting to remove
political considerations from cost-benefit analysis whenever possible, replacing it with
sound methods founded in international literature that can then be viewed through a
political lens. The following explores each of the six areas with an eye towards the existing
New Zealand context, pitting it alongside the best literature in the area.
Section 2.01 The Discount Rate
Discounting practice is easily described. It is a tool used to represent the value of a
future dollar in today’s terms. The methods used to achieve this conflict are primarily
divided into two camps. One originates in the private sector, financially-oriented
2.01—7
discounting, and primarily uses market rates of interest with a risk premium to discount
future values. The second, social discounting, attempts to discount at the rate of preferred
consumption, rather than the rate of private interest, to reflect the concept’s psychological
underpinnings.
Discounting is a relatively recent concept in government, though it dates from far
before in the investment world. After making an appearance in government analyses in
the 1970s, the 1980s saw discounting rise to prominence (Sáez & Requena, 2007),
revolutionizing the practice of cost-benefit investment analysis by incorporating the
practical challenges imposed by the cost of capital and the inherent individualistic desire
to consume now rather than later. This has resulted in a modern status quo where CBA
influences every government investment in the developed world to an extent that the
OECD notes “The discount rate is normally the most crucial factor in whether medium
to long-term projects pass a cost-benefit analysis” (2007, p. 5). Given this fact, both the
why and how of discounting need be examined when looking at medium-term
investments, among which transportation fits quite well.
Various writers propose different explanations for discounting as a practice. Some
see it employed to represent the quantifiable chance that the population intended to enjoy
it will not be there, or the investment will not be there to serve them, due to some manner
of catastrophe (Feldstein, 1964; Rambaud & Torrecillas, 2005). Others see the social
time preference to consume now rather than later as the primary focus (De Rus &
Nombela, 2007; Feldstein, 1964; Hope, 2008; OECD, 2007; Rambaud & Torrecillas,
2005), while still more, including New Zealand, see the defining point of view as market
rates of return layered upon the cost of capital (Boardman, Moore, & Vining, 2010;
Lally, 2012; New Zealand Transport Agency, 2010; OECD, 2007; Rambaud &
Torrecillas, 2005). At a fundamental level, there is compelling reason behind each of
these options. For the purpose of transport investment, the existing literature’s point of
view on discount rate boils down to a few practical options:
• Using a fixed discount rate based upon the cost of capital and market return.
• Using a low, fixed discount rate that reflects respect for future generations’ needs.
• Using declining discount rates that track the long-term social perception of value.
• Refusing to discount at all, based on the responsibility to future generations.
The current New Zealand Treasury guidance on transport investment discounting
is six percent – just reviewed in July 2013 – due primarily to the logic of the first option
above (New Zealand Transport Agency, 2010). Market return on investment is
considered and strong arguments may be put forth that the cost of capital alone justifies
the use of this method in government evaluation (Lally, 2012). One of the key elements
to consider, however, is the broad base of literature that recognizes such discount rates
fail to perform adequately both in public and private arenas. This literature ranges from
that dealing with forestry (Hepburn & Koundouri, 2007; OECD, 2007), to that dealing
with broader ecological issues (Hepburn, Koundouri, Panopoulou, & Pantelidis, 2009;
Hope, 2008; Rambaud & Torrecillas, 2005) and generalized public investment (OECD,
2.01—8
2007; Rambaud & Torrecillas, 2005; Sáez & Requena, 2007). The key factor these works
acknowledge is that such discount rates as envisioned currently do not take into account
several important issues.
Getting to the source of these issues begins with uncertainty and aggregate
preferences. Gollier (2002b) details that discount rates as currently employed don’t
account for the uncertainty surrounding future growth, and thus “…the discount rate to
be used for long-lasting investments should be a decreasing function of their duration.
This is due to the negative effect of accumulating the per period growth risk in the long
run” (p. 163). Further research of his bolsters this position with models that deal with
other areas of discounting theory. He models the interaction of wealth effects and
precautionary action (Gollier, 2002b) and aggregate time preferences given non-
heterogeneous rates of impatience among a population (Gollier & Zeckhauser, 2005)
which separately lead to the same conclusion that fixed discount rates do not accurately
reflect social time preference. These combined models lead from different avenues to the
same conclusion that declining discount rates should be employed by any agency
evaluating an investment. The conclusion that discount rates should be declining no
matter what the rate is to begin with has an outsized influence on the outcome of cost-
benefit analysis, particularly in the long term. Rambaud (2005), OECD (2007), Hepburn
(2007 & 2009), Boardman (2010), and Azfar (1999) all note these issues with a constant
market-focused discount rate.
Weitzman (1998) further highlights the problems with the existing regime by
pointing out practically that, “the logic of exponential discounting forces us to say that
what we might otherwise conceptualize as monumental events ‘do not much matter’
when they occur in future centuries or millennia… Yet almost no one really feels this way
about the distant future”(p. 201). His modeling both adds further ammunition to the
conclusions posed by Gollier, and provides an additional conclusion that certainty-
equivalent social discount rates are the most appropriate rates to employ when looking at
the distant future. By modelling future uncertainty about growth rates, Weitzman shows
that, in fact, future growth rate uncertainty need also be accounted for when factoring
risk into a cost-benefit analysis. Project risk is offset, or more than offset, by the
fundamental risk of growth rate uncertainty. His modeling leads to the conclusion that
the least limiting discount factor should be employed for any given evaluation (Weitzman,
1998). The least limiting discount factor is the lowest one, particularly as you get further
into the future. As a result, he finds that the lowest discount factor possible, and thus
discount rate possible, need be employed. In the near future, the lowest possible discount
rate is the certainty-equivalent discount rate. In the distant future, it is more likely to be
some diminished portion of that rate.
In sum, the discount rate currently employed by the NZTA EEM under option
one is denying society a proper picture, and in turn possibly introducing a cost by
selecting projects that are not optimal among projects being evaluated. This is
particularly true in the transport space, where investments tend to be long lived.
The second option, a low long-term discount rate as proposed in the Stern Review
(Stern, 2006) for environmental purposes, offers value for intergenerational purposes. In
2.01—9
particular when applied to investments or impacts that aren’t necessarily consumption-
driven, and that will be felt long after the existing generation has passed, this method has
high potential utility. That said, it is particularly unsuited to the evaluation of near-term
government action due to its neglect of short-term project risk. This flaw is markedly
similar to the existing regime’s flaw, but with the opposite outcome. It is because of these
opposed strengths and weaknesses that some literature notes the possibility of applying
different discount rates to different impacts (Rambaud & Torrecillas, 2005) – for example
discounting environmental impacts less than financial revenues. These possibilities are
posed, and summarily declined due to the possibility of lending overwhelming effect to
certain categories in evaluation. Sáez & Requena (2007) offer a particularly illuminating
taxonomy of discounting regimes and the relative merits of each.
Based on the understanding that a change is warranted due to these shortcomings,
it is important to note that it is easy to eliminate the fourth discounting option – not
discounting at all. As mentioned previously, the risk of the project or the population
served being impacted by catastrophe, thus preventing the future use of any good, should
set a baseline minimum for discounting. That minimum is quite low, and it resembles
closely the second option.
The combination of these strengths and weaknesses of options one, two, and four
suggest that option three, a declining fixed (or hyperbolic) discount rate that applies to all
government investments equally, is the optimal discounting scheme to employ (Azfar,
1999; Gollier & Zeckhauser, 2005; Gollier, 2002a, 2002b; OECD, 2007; Rambaud &
Torrecillas, 2005; Sáez & Requena, 2007; Weitzman, 1998). Schemes based on this
concept have already been adopted by the United Kingdom and France, with the United
States scheme bearing a remarkable resemblance (Her Majesty’s Treasury, 2013;
OECD, 2007; Rambaud & Torrecillas, 2005). A number of countries bear similar
resemblance when noting the differential discount rates employed in different areas of
government. The United States highlights this policy, with financially-oriented (Office of
Management and Budget) government departments tending towards higher rates, and
environmentally- or infrastructure-oriented (Environmental Protection
Agency/Department Of Transportation) government departments tending towards lower
rates (OECD, 2007; Rambaud & Torrecillas, 2005). This ultimately is linked to the
United States’ downward-sloping yield curve. While employing differential rates, these
countries would be better served by a unified declining discounting scheme that was
employed across the whole of government, whether hyperbolic or fixed, but declining.
Due to this growing global consensus, and the state of affairs in New Zealand,
adopting a similar program would yield benefits for the capability of NZTA and other
New Zealand entities evaluating future projects, including transport investments. This is a
revision already adopted in several OECD countries, highlighted by France and the UK.
Below is the framework of their discounting schemes, set side by side with a pair of
smoothed versions to combine the best of both in creating one for New Zealand.
Table II-1 Existing International & Proposed NZ Discounting Schemes
Year of Project UK France NZ Social NZ Compromise
2.01—10
0 – 30 3.5% 4% 3% 4%
31 – 75 3% 2% 2% 2.5%
76 – 125 2.5% 2% 1.5% 2.25%
126 – 200 2% 2% 1% 2%
201 – 300 1.5% 2% 1% 1.5%
301 – 1% 2% 1% 1%
This discounting scheme strikes a balance between the two existing discounting
schemes, acknowledging lower annual growth rates per capita in New Zealand (Institute
for Health Metrics and Evaluation, 2013) compared to both the United Kingdom and
France, and smoothing together the existing schemes. This scheme offers advantages for
reasons detailed above, but also because the focus shifts to the longer term as long-term
interest rates reduce. This removes the focus on shorter-term analysis which can be more
variable as it is affected by fluctuating interest rates, and shifts it to the long-term growth
of the economy. This social discounting regime may be challenged either due to
institutional inertia or pre-existing notions of the political implications of such a
discounting regime. This paper does not concern itself with political issues, but does
propose that a compromise discounting regime such as proposed above might provide a
middle ground for social discounting and market rate discounting to settle upon. This
regime uses short-term certainty-equivalent rates in the 30 year timespan and smoothed
decline over time. This compromise reflects the different environment and solidifying
consensus that declining CBA is appropriate. For the purposes of our evaluation, this
paper uses the social discounting schemes.
The above scheme yields two technical challenges that would need to be
addressed by the EEM as well. Firstly, the period of evaluation should be extended
beyond 10-30 years, to a minimum of 50 for any project – and ideally 100. Pragmatically,
using the existing discount rate negates the value of any such evaluation – but this scheme
would require closer attention to long-term trends.
Secondly, due to the detachment of market rates and the social discount rate, it
would be necessary to discount payments against project financing as well. Currently, that
is not necessary because certainty in financing payments implies the interest rate and
discount rate are the same due to the market basis, and therefore discounting nets out in
the original value of the financing. This would not be the case if the discount rate and
market rates are detached in the manner described, and thus requires such evaluation on
a project level.
It is worthy of note that costs should be discounted based on when the resource
cost is incurred. It also serves a second purpose of ensuring interest rates are accounted
for in an environment where the discount rate is detached from the market rate of
interest. In this way, payments account for existing market interest rates, yet reflect the
2.01—11
overall net present value of the financing payment’s impact. This distribution of impact
better reflects the potential for benefits, while implying that separate and severable all-of-
government considerations but be evaluated for how much debt is tolerable outside of the
cost benefit analysis process.
Section 2.02 The Value of Time
The value of time for cost-benefit analysis is explained relatively easily. It is used
to calculate the “cost” of a given trip in more than financial terms. This means,
effectively, that a higher value of time equates to a more troublesome commute. This is
why modes with co-benefits (public transport, which allows working and reading, or
cycling which has health benefits) tend to have a lower value of time than driving, which
is perceived to be the most problematic mode.
Quantifying the value of time (absolute time saved) and reliability (reduced
variability) is one of the oldest challenges for cost benefit analysis when analyzing
transportation investments. This is not just because valuing time is difficult due to its non-
economic nature, but also because times vary based upon both work purpose and travel
mode. O’Fallon & Wallis (2012) examine the reasons behind this and the existing
attitudes in New Zealand. Their research includes such insights as walking time is valued
more as leisure, even when commuting, while driving is valued more like work. The
NZTA distinguished between 9 separate values of time for different modal choices and 3
separate types of travel for analysis (New Zealand Transport Agency, 2010) until July
2013, when it removed the modal choice distinctions. The variance in value between
modes employed matched well with international literature. The values themselves for the
three types of travel (during work hours, to and from work, and leisure/optional travel) on
the other hand do not, except for leisure travel. Current New Zealand numbers indicate
roughly $22/hour for travel during work hours, around $7.50/hour for commuting to
and from work or leisure time travel (New Zealand Transport Agency, 2010). This
section explores the commuting value of time because the number of trips taken for
commuting purposes dwarfs the number of trips taken during work hours.
Various theoretical studies have posited that the value of time is in fact quite low,
particularly in situations where “within-route” times are in consideration (Hopkinson &
Wardman, 1996). The evidence base for this is highlighted by studies built upon
willingness-to-pay foundations such as Calfee & Winston (1998), who find through stated
preference methods the value of time to be in the vicinity of USD$4/hr. Within a New
Zealand context, a stated preference study of Auckland drivers found the value of time
between $7.54 and $20.7 per hour depending on travel conditions (Hensher, 2001). This
is not too far distant from the New Zealand EEM value of time of NZ$7.40/hr
determined using similar stated preference methods.
On the other hand, studies using revealed preference following toll payments to
try and value these aspects tend to show much higher values of time, along with an even
more noticeable increase in value of reliability as measured by reduced variance in travel
times from day to day (Brownstone & Small, 2005; Lam & Small, 2001). Women in
particular find reliability 40% more valuable than time (Brownstone & Small, 2005).
Keep in mind that reliability is calculated as the variability in travel time, which by and
large is much smaller than overall commute time. As a result, the value of a single minute
2.02—12
of variability is far higher, but the magnitude of the overall effect of reliability on a
commuter’s cost structure for commuting is much lower.
The literature dissects the values relative to the prevailing wage and gross wage
rate, leading to the conclusion that the value of travel time itself varies between 20% and
100% of the prevailing wage, depending on congestion, with the strongest argument
being for 72% and above when attempting to divine an average for a congested zone
(Boardman et al., 2010; Lam & Small, 2001; Litman, 2013a). It is notable that O’Fallon
& Wallis (2012) quote a conference paper as disproving the link between prevailing wage
and travel time value. However, later work by the same author has in fact proved the
link(Jara-Diaz & Guevara, 2003). In addition, the bulk of research indicates such a link
does exist, and in fact wage-based tools are employed by the United States Department of
Transportation, Canada and much of Europe. Given the large body of work supporting it
and the mixed body of work from that author, the O’Fallon & Wallis, (2012) assertion
must be disregarded.
With a prevailing wage of NZD$31.90 and choosing 72% as a point to use,
Wellington resembles closely the data Lam & Small's (2001) optimized model used as
input data, though Lam and Small used US Dollars. Their sample indicated a prevailing
wage of USD$31.69. This indicates a value of time, both for Lam and Small and for
Wellington, in the vicinity of $22.87. Despite the different currencies, the values are for
practical purposes the same. This high value is supported by recent research at the New
Zealand Institute for the Study of Competition and Regulation using hedonic price
models that indicated a value for each minute of travel time to the central city of around
NZD$40/hr spread out temporally over the span of ownership for a house (Pettit,
Daglish, Saglam, De Roiste, & Law, 2013; Wannan, 2013). 1
The value of reliability then layers upon this at 66% of the value of time for men,
and 140% the value of time for women (Brownstone & Small, 2005; Lam & Small, 2001).
Rather than conducting gender sampling an evaluation can instead use 100% of the value
of time to calculate the value of reliability for cost-benefit analysis purposes.
Employing these values as baseline, there are three major influencers needing
consideration for the value of time indicated in the literature:
• Purpose of travel – non-optional travel like commuting is burdensome and
time is valued higher when computing costs than for optional driving.
(Handy, Weston, & Mokhtarian, 2005; Litman, 2007)
• Mode of travel – commuters pay substantially more to avoid driving than
to avoid other modes, which implies that they find driving more expensive,
or other modes offer more co-benefits. (Elvik, 2000; Litman, 2007, 2009;
Russell, 2012)
1 This research indicated a home value increase of NZD$6700 for each minute of public
transport travel time closer to Cuba Mall a home was. When accounting for estimated
number of commuting minutes per year this amounts to around NZD$0.67 per minute
value of time, or NZD$40.20 per hour.
2.02—13
• Reliability of travel – consistency of schedule, minimization of wait times
and congestion. (Handy et al., 2005; Lam & Small, 2001; Litman, 2009;
Russell et al., 2011)
The EEM adequately categorizes different modal choices and purposes of travel,
with three categories of purpose (work, commute and other) and 10 categories of modal
choice (driver and passenger for car/motorcycle, light freight, heavy freight, seated or
standing bus, pedestrian and cyclist). As of July 2013, it has now eliminated the modal
differences, however, introducing a new wrinkle to the EEM’s challenges. Even aside
from this, the EEM’s problems in its treatment of travel time are still noticable.
Incongruous differentials between purpose and modal split are difficult to justify in
light of the existing literature. Particularly glaring is the differential between the roughly
equal during work hours travel time values (which match closely with the value this paper
determined above for commuting values) and the travel time values for getting to and
from work, which are roughly equivalent to the values used to represent “all other” travel
modes (New Zealand Transport Agency, 2010). This does not conform with the findings
of Handy (2005), Russell (2011,2012), Lam (2001), Litman (2009), or De Rus (2007).
Though none of these specify a value of time for non-commute purposes, the consensus is
clear that the value of time for commuting purposes is roughly equivalent to the values
NZTA EEM specifies for at-work travel.
Possible explanations for this incongruity are a presumed differential based on
time being paid or business savings being prioritized in evaluations. However, rather than
devaluing accurate measures of travel time’s value for commuting, the value for at-work
travel should be increased above the commute baseline. As such, values for at-work
should roughly equate to the value of time plus some representation of the cost to business
for having their employees idled during that time, while commute time value should sync
up with the above-determined value of time. The easiest way to represent this would be to
add the prevailing wage atop the isolated commute costs. Optional travel can be valued at
less than commute levels, but the current EEM level conforms with research that
indicates lower values of time for costing optional travel. Further research is needed to
determine both numbers for during work hours commute time and leisure travel. This is
particularly true of leisure travel. With leisure travel it is possible per the work done by
Handy (2005) that additional optional travel time may actually add value to an
individual’s enjoyment, rather than being solely viewed as a cost, providing wide-ranging
possibilities for values.
An additional issue with the EEM is that the way it calculates congestion and
bottleneck penalties. These calculations use a value far below even the travel time
numbers denoted in the optional travel section of time values, roughly NZD$4 per
veh/hr (New Zealand Transport Agency, 2010). It neglects the possibility of multiple
passengers and establishes that value of time in congestion is worth far less than when
travelling at speed. This disagrees both with stylized facts that driver frustration due to
congestion would incur a higher cost than actually travelling towards the destination, and
with stylized facts that regardless of whether travelling at speed or waiting in traffic, the
commuter’s time is worth the same. Additionally, congestion value calculations focus on
an incredibly complex section devoted to determining exactly how to quantify the vehicle
2.02—14
hours of congestion caused by intersections, exceeding roadway capacity, and bottlenecks
in general. This method generates an artificially low value of congestion costs. This
should be abandoned in favor of valuing congestion equally to travel time and doing so
on a per-person rather than per-vehicle basis. Such an evaluation would create an
underestimate as well, as frustration induces a perceived cost. However, it would be much
more accurate to reality than the existing system.
Layered atop this is the reality that reliability of transport is viewed, on average, as
equally or more important to the value of travel time itself (Brownstone & Small, 2005;
Lam & Small, 2001). This is particularly true of women (Lam & Small, 2001), who have
a slight population advantage over men in New Zealand and elsewhere, thus raising the
value of reliability (World Bank, 2013). That said, reliability accounts for roughly one
third the service quality variance compared to travel time value representing about two
thirds of the variance value (Brownstone & Small, 2005) when comparing express lanes
to standard traffic patterns. To clarify, given a comparison between any two route
choices, any differences between them in value of travel quality would be apportioned by
the above ratio. This ratio may not hold when compared to the more regimented
schedules of a train system, wherein reliability is highest, but it provides a useful indicator
of what that higher value of reliability equates to in car-to-car comparisons or bus-to-train
comparisons, at least. The higher value for time value and the higher multiple implied by
the literature combine to raise the value of reliability using the EEM formulas by at least
50%.
There is one existing issue with the EEM formulas, which rely heavily on capacity
and other engineering statistics, in that the formulas don’t explicitly demand that
population increases and development changes be taken into account for future years. For
example, due to the widely-acknowledged concept of induced demand lane expansions
inspire development further down a given highway, usually inspiring in the medium term
a higher capacity overuse than existed before (Goodwin, 1996). This should be accounted
for in long-term analyses. Note that this is true of train systems as well, but no meaningful
intraurban rail systems exist in Wellington (or even New Zealand) to introduce such an
issue.
Finally, the EEM also neglects to present different values for time when cycling or
using any public or active transport mode. Cycling and public transport offer co-benefits
in terms of health, decreased need of gym fees, productivity benefits, increased leisure
reading time and other benefits detailed in Hopkinson & Wardman (1996), Russell
(2012), Russell et al. (2011), and Sælensminde (2004). What this amounts to is a lower
value of time for costing purposes when using public transport, and lower overall costs
when cycling or walking, roughly half that used when travelling by car. This is
represented in the older EEM as time is worth NZD$7.40 when commuting by car, and
only NZD$4.80 by public transport (New Zealand Transport Agency, 2010). However,
this was revised in July 2013 to equal out in spite of the literature’s evidence to the
contrary. The evaluation of this modal difference is beyond the scope of this paper,
however it is a ripe topic for further primary research in New Zealand. For the purposes
of our corrected costs, due to the research consensus on this matter this paper encourages
further research to determine the value of the benefits accrued on a per hour basis for
active and public transport, as well as psychosocial benefits for car travel, to then apply
2.02—15
against this value of time. This paper operates as if the status quo of unequal travel times
is accurate – using a 64% multiplier (4.80 divided by 7.40) to acknowledging that public
and active transport travel time is less of a nuisance than driving travel time.
Section 2.03 Treatment of Cycling Benefits
Carrying on from this challenge, cycling, as noted previously, has co-benefits in
the health area. Valuation of these benefits is typically done on a per-km basis to
conceptualize both the dangers of cycling in certain contexts (a cost), and the value of the
health benefits accrued (a benefit). These are notably flawed in the New Zealand
Transport Agency’s EEM.
Cycling trips have experienced a massive decline in New Zealand over recent
decades, and evidence suggests this to be a result of transport policies that fail to
adequately account for the needs of cyclists (Jakob, Craig, & Fisher, 2006). The 2010
EEM makes strides to rectify this, but is hamstrung by several issues that distort the value
of cycling in the eyes of evaluators. Currently, the methods focus too heavily on travel
time, and neglect that safety is more important to cyclists than any other consideration
(de Geus, De Bourdeaudhuij, Jannes, & Meeusen, 2007; Hopkinson & Wardman, 1996;
Sælensminde, 2004). As a result, cost-benefit analyses of active transport undervalue the
total social benefit available to society if planners create cycling routes that are both safe
and fast, rather than just fast. Having already discussed the challenges implicit in the
evaluation of the value of time when cycling, it is necessary to look at other aspects of the
EEM’s treatment of cycling.
The absence of risk reduction from benefit consideration when investing in
cycling-friendly infrastructure is a prominent, and surprising, oversight considering the
overall vulnerability of cyclists compared to other commuters and the research
prioritizing safety (New Zealand Transport Agency, 2010). This is particularly strange in
view of the fact that the only other project type that does not consider risk reduction is
land use and parking investments, and the EEM lays out the point that “cycling and
walking improvements should specifically address safety and personal security issues”
(New Zealand Transport Agency, 2010).
Higher road speed and proximity to traffic or parking alongside roads (due to the
danger of opening doors) leads presumably to higher feelings of insecurity, which should
be accounted for in the methodology. Measuring safety perception in the EEM is
captured separately by willingness-to-pay methodology to capture all benefits. Currently,
willingness-to-pay assigns a value of $1.30 per km for health benefits and $.05 per km for
improved safety (New Zealand Transport Agency, 2010), which does not comport with
research indicating that security is more important than any other consideration to
cyclists (de Geus et al., 2007; Sælensminde, 2004).
Safety and risk reduction (security) is prioritized by cyclists above all other
considerations, and its value can be as much as the value of travel time – implying that it
is perceived as a requirement rather than an option to employ cycling as a commute
mode (Elvik, 2000; Hopkinson & Wardman, 1996; Sælensminde, 2004). Despite this
2.03—16
lesser value, it is important to get this particular aspect of evaluation correct considering
the top-of-mind nature in the mode choice decision. To represent this, Sælensminde
(2004) notes that Nordic literature indicates a value of NOK 2/km (NZD$.44/km). This
paper cannot access this literature as there is no English translation. However, based
upon the research that indicates insecurity is even more important than the marginal
travel time cost of cycling, it would seem that this value is somewhat conservative, and
thus sensible to employ for low-speed areas until research on the value of insecurity can
be conducted more extensively in New Zealand.
Another alternative is to set the value of time ($22.87) equal to the value of
insecurity, dividing by the travel speed of commuter cyclists (14km/hr) yielding a
somewhat higher value of insecurity of NZD$1.63/km. This likely sets a fair upper bound
for the value of insecurity. That said, given the context where insecurity is perceived to be
the most prominent impediment to cycling according to international research, it is likely
that this value would be more productive to employ to accurately represent the
psychosocial and perceptual benefits of safety. Numbers closer to the upper bound should
be adopted for high-speed areas (over 30km/h speed limit) and closer to the lower bound
for low-speed areas (15km/hr and below).
We can draw the conclusion that the EEM either neglects to separate out
improved security, or fails to apply adequate value to the issue. Some would indicate this
is the responsibility of the WTP surveys, which has some validity. The health benefits are
positive externalities, and though they have higher economic value, they are intrinsically
valued less by cyclists when they make the cycling decision. This is because these benefits
are accrued primarily to the government-funded healthcare system as a positive
externality, while the individual benefit represents very little of that excepting intangible
costs like quality of life. Because of this, the values generated by WTP stated preference
methodology for transport purposes are brought into question. The issue with the EEM’s
treatment of cycling extends into the calculation of health benefits. The EEM poses:
“Walking and cycling can have significant health benefits through increased exercise levels. However, this
could be offset by an increased exposure to pollutants if the activity involves sharing road space” (p. 2-30).
However, research shows that it is not cyclists who endure the highest levels of
pollutant exposure, but in fact motorists and diesel bus passengers who experience the
pollutants in an extended fashion (Kingham, Shrestha, Longley, & Salmond, 2011).
Contrasting this, electric light rail is the only mode that is superior to cycling in terms of
total pollutant exposure, though cyclists do experience higher peak intensities for
extremely brief moments. The value of health benefits at $1.30 per km presumably
includes a penalty for pollution exposure, but it is unknown to what extent. To compare,
the World Health Organization HEAT (Health Economic Assessment Tool) indicates
that an appropriate value for cycling per km would be $3.5/km using updated values of a
statistical life from NZIER (2010) (World Health Organization / Europe, 2013). Given
the lack of transparency for the prior number and the global applicability of the HEAT
tool, a value closer to the latter should be adopted.
2.03—17
In sum, numerous changes are needed to reflect current literature in the way
cycling is treated for cost-benefit analysis in New Zealand. The changes are summarized
in Table II-2.
Table II-2 Proposed Changes to Cycling Benefits
Area of Change Existing Value New Value
Value of $.05/km $.44/k, to $1.63/km
Insecurity/Safety
Value of health $1.30/km $3.50/km
benefits
One thing is made clear. Whatever individual benefits are created by cycling are currently
more than offset by any insecurity disbenefits that exist. This is particularly true if one
notes that existing benefits in total can be more than offset by the best practice value for
insecurity alone. By removing the value of insecurity from play using segregated
cycleways, much more accurate data on the value of cycling’s individual benefits could be
generated, and it seems the government health system could accrue an abundance of
potential health benefits were that the case.
Though these represent significant changes to bring the EEM up to the state of the
art, for the purpose of later evaluating the Spine Study they have little effect excepting
when cyclists transfer to public transport as their mode of choice. Because there are so
few cyclists in Wellington, however, and the value of time benefits of switching to public
transport may well outweigh the value of health benefits, this paper does not calculate
this. That does not in any way minimize their importance when looking at cycling
infrastructure in particular.
Section 2.04 Patterns of Development and Transport-Inspired
Value
Patterns of development, and specifically property value influences, are
inadequately handled in the EEM. It is well-known that transport influences property
value. These value increase tend to “capitalize” the itemized benefits – in that they
indicate what the sum of total individual benefits are worth to those who wish to use the
transport. When they buy a house, or when they buy a business in the area what extra
customer value they can expect. Currently, these are ignored as “double counting” in the
EEM, but they offer the opportunity for good values to “check” that the benefits
indicated by a cost-benefit analysis are sensible.
There are two major areas where the EEM neglects important aspects of transport
development, and development patterns and land value changes are one of them. The
second is private automobile costs. This paper first addresses development patterns. The
EEM states that:
2.04—18
“Certain external impacts of activities, such as increased land values, may arise
because of the improved level of service and accessibility to nearby areas. These
impacts shall be excluded from the evaluation because including them would be
double counting” (New Zealand Transport Agency, 2010, p. 2-6).
While widely accepted, problems with this point of view have become evident in
recent years. Improved level of service and accessibility are tangibly different from the
value intrinsic to the improvement of the area surrounding a station. For example higher
foot traffic surrounding stations drives superior business opportunities. The superior
business opportunities thusly drive better businesses, which in turn promote easy
shopping for local residents. But extending further, anything from the design of any
installed vehicle or the attractiveness of stations can affect land values. While a certain
percentage of capitalized land value certainly represents typified benefits, these typical
benefits do not account for the full scope of benefits. Improved travel time and reliability,
for example, represent only a small percentage (.000001%) of the capitalized benefits
potential predicted in the Spine Study.
Practitioner workshops have noted this contradiction in recent years
(Transportation Economics Center, 2010), providing evidence that:
“While great care should be taken to avoid double counting, recent evidence
indicates that part of the increase in land value may be more than just the
capitalized value of the transportation and accessibility benefits. It may include
option value accrued by non-users of the transportation facility” (p. 18).
The most convincing argument that including some form of capitalized land
value is not double counting comes from literature that details the difference in hedonic
price effects between modes of transport. Meta-analysis of commuter services price effects
finds that all else being equal, bus rapid transit stations have no significant effect on land
values, while rail stations cause highly significant increases based on proximity
(Debrezion, Pels, & Rietveld, 2007). But what this means is that, in effect, rail modes have
intrinsic value uplift based on characteristics not captured in cost-benefit analysis or the
EEM.
It is important to ensure that, particularly in densely-populated areas where
properties are packed in tightly, land-value changes are accurately captured as the
influence could be enormous. Perhaps Lewis-Workman & Brod, (1997) said it best when
they noted:
“The presence of transit contributes to the character and form of neighborhoods and
creates opportunities whose value may not be fully captured by the intensity of use of the
transit system” (p. 147).
It is that contribution to urban form that needs to be better captured in the EEM.
To attribute the capitalized cost to benefits, it is important to state what attributes drive
this value. When including choices like simplifying the car ownership decision, increasing
option value, and allowing all valid transport decisions equal opportunity, the value is
evident. Perhaps less evident are the minor differences in reliability and character, such as
2.04—19
You can also read
